Optimal Timing for Stock Trading in USA: Data Driven Signals for Buy and Sell

Madireddy, Girija (2023) Optimal Timing for Stock Trading in USA: Data Driven Signals for Buy and Sell. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (5MB) | Preview
Preview	PDF (Configuration Manual) Download (3MB) | Preview

Abstract

Stock market forecasting is an interesting topic since there are a lot of variables that can affect prices in the future, along with unanticipated noise. Still, the ability to analyse stock market trends could be vivacious to researchers and investors, and thus has been of continued attention. It is tremendously imperious to distinguish the equilibrium between financial perceptions and econometrics, learning tactics, computational astuteness, technical indicators, and their assimilation in order to analyze and study the arena of interest. In this research, graph theory is introduced as an approach. This method uses Spatio-temporal relationship data among diverse stocks by modelling the stock market as a composite network. Long Short-Term memory networks is combined with this graph-based approach to form a hybrid model. Graph Convolutional Neural networks (GCNs) are recognized for their capacity to exploit Spatio-temporal relationships among diverse stocks, while LSTMs excel in handling sequential time series data for prediction modelling. An experiment had been conducted to add sentiment analysis of daily news headlines along with stock price data as input to the model. These deep learning models are equated in contrary to a conventional statistical time series model to assess both computational efficiency and prediction accuracy. This research work concluded that GCN-LSTM model without sentiment analysis is performing better in predicting individual companies stock movement, while the method of aggregation needs to be updated to weighted voting instead of majority voting when calculating overall DJIA movement. Deep learning models are performing far better when compared with statistical ARIMA model since they leverage structural and temporal information from data.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Staikopoulos, Athanasios UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science H Social Sciences > HG Finance > Investment Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HG Finance > Investment > Stock Exchange
Divisions:	School of Computing > Higher Diploma in Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	16 May 2025 10:07
Last Modified:	16 May 2025 10:07
URI:	https://norma.ncirl.ie/id/eprint/7560

Actions (login required)

View Item